We’re announcing today the third batch of Open Source AI Grant recipients.
As a reminder, this program gives grant funding (not an investment) to hackers, researchers, and small teams doing important work to support AI development outside of the major labs. You can see the prior recipients here and here.
This latest batch focuses on better understanding the capabilities of frontier LLMs and testing their boundaries. Projects like SWE-Bench and ARC Prize provide essential frameworks for measuring progress, moving beyond saturated benchmarks to test genuine reasoning and real-world problem-solving. Experimental work from researchers like truth_terminal, Pliny, and Janus pushes the limit of what’s possible with AI, often revealing unexpected capabilities and raising important questions about how AI interacts with the “real” world. And as always, we’re giving grants to several of the top AI infra teams working on open source projects.
The full list of recipients is:
Thank you to all our grant recipients for your contributions to the field. Your work ensures that the future of AI remains open, innovative, and accessible to all.